Problem with used disk space on HD Space root

[fluke]

Hi,
I am having problem identifying what has filled my disk space on HD Space root.
Proxmox and 'df' report 88% used space:

proxmox1:~# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/pve-root   95G   79G   12G  88% /
tmpfs                 7.9G     0  7.9G   0% /lib/init/rw
udev                   10M  768K  9.3M   8% /dev
tmpfs                 7.9G     0  7.9G   0% /dev/shm
/dev/mapper/pve-data  197G  8.4G  189G   5% /var/lib/vz
/dev/sda1             504M   67M  412M  14% /boot
//172.16.101.200/VMBackups
                      2.0T  265G  1.7T  14% /mnt/pve/winshare
172.16.101.121:/VMBackups
                      839G   77G  762G  10% /mnt/pve/VMBackups-1

But 'du' reports, what I think is more realistic:

proxmox1:~# du -hxs /
1.2G    /

I have read, that this inconsistency might be due to the “open file descriptor” problem - file is deleted, while still in use. This is solved, when the process, which keeps the file in use is killed. But on my machine I could not identify such process:

proxmox1:~# lsof|grep deleted
apache2    5383     root   13w      REG              252,1          0    1590760 /var/run/apache2/ssl_mutex (deleted)
apache2   19590 www-data   13w      REG              252,1          0    1590760 /var/run/apache2/ssl_mutex (deleted)

Anyone has an idea what uses my 78GB. I am afraid, that my backups might fail due to lack of space.

Hint: 5 days ago I had failed backups, maybe vzdump left something behind?

 

[fluke]

I managed to figure out the problem. I am using samba share for backup, but when for some reason mount fails, backups are written to local disk. Later when mount succeeds, it hides these local backups. Anyone has an idea how to prevent local backups when samba mount fails?